AgentOp
AgentOp lets you build AI agents that run entirely in your browser. Open models like Llama, Qwen, Gemma, and Phi load locally with WebGPU. No install, no server, no API key, and your prompts never leave the tab. You write your agent's logic in Python, test it instantly against a local model, then export everything as a single standalone HTML file including the model loader, code, UI, and encryption.
Gallery
FAQ
AgentOp lets you build AI agents that run entirely in your browser. Open models like Llama, Qwen, Gemma, and Phi load locally with WebGPU. No install, no server, no API key, and your prompts never leave the tab. You write your agent's logic in Python, test it instantly against a local model, then export everything as a single standalone HTML file including the model loader, code, UI, and encryption.
Discussion
AgentOp looks genuinely useful — the AI Tools space needed this. qq: does it scale cleanly for bigger teams, or is there a performance hit to watch out for? Congrats on shipping 👏
@vikram great question — it stays light even on large workspaces; nothing extra to configure. Appreciate you checking it out!
@ozgun great to know, appreciate the transparency 🙌
The AI Development angle is the part that got me — feels way cleaner than what's out there. For an existing setup, how granular can you get about what you expose? Congrats on shipping.
@sanjay thank you! you can scope it as tightly as you want — per project, per workspace, your call.